AI Connect Service - Service Specifications

Download OpenAPI specification:Download

Service Specifications

The iotcomms.io AI Connect Service provides robust and flexible integration of voice applications with AI platforms, enabling advanced analytics and conversational AI capabilities. Designed for seamless interoperability with Amazon Web Services (AWS), Google Cloud services, and custom application servers, the AI Connect Service abstracts telecom complexities and supplies audio and call metadata for processing by the connected application or AI platform.

  1. Separate Media Streams for Participants

    Each call participant is sent as an independent media stream, regardless of the integration type. This enables granular analysis of individual interactions in real-time or post-call scenarios. Whether working with AWS, Google, or custom application servers, this approach ensures precise handling of participant-specific audio and metadata for enhanced accuracy in analytics and conversational AI.

  2. Real-Time Events for Call Metadata

    The service provides real-time event notifications containing detailed call metadata to enable advanced analytics and decision-making. This data includes:

    • Calling Party Number: The phone number of the caller.

    • Called Party Number: The phone number of the recipient.

    • Network Routing Information: Details about the routing of the call through the network.

    • IP Addresses: The source and destination IP addresses involved in the call session.

    • Session Description Information (SDP): Details about the media streams, codecs, and connection parameters negotiated during the call setup.

  3. Integration with AWS Services

    The AI Connect Service integrates directly with key AWS AI offerings to enable voice analytics and conversational capabilities:

    • Amazon Transcribe

      • Integration: The service uses Amazon Transcribe's API to convert voice streams into real-time transcriptions.
      • Supported Features: Includes speaker diarization, custom vocabularies, and language identification for precise transcription in diverse use cases.
      • Usage: Real-time and batch transcription workflows are supported, with output formats directly compatible with analytics pipelines.

    • Amazon Lex

      • Integration: Connects with Amazon Lex to enable natural language understanding (NLU) and conversational interfaces.
      • Supported Features: Supports multi-turn dialogues, slot filling, and contextual intent recognition.
      • Usage: Facilitates building AI-driven voicebots and chatbots for automating customer interactions.

    • Security and Compliance

      • Communication between the AI Connect Service and AWS services is secured using AWS Identity and Access Management (IAM) roles and policies.
      • AWS-hosted data complies with HIPAA, GDPR, and other regulatory standards.

  4. Integration with Google Cloud Services

    The AI Connect Service integrates with Google Cloud's AI tools for voice interaction and data analysis:

    • Google Dialogflow

      • Integration: Dialogflow APIs are used to process text inputs from voice transcriptions and enable conversational AI workflows.

      • Supported Features: Includes prebuilt intents, multi-language support, and context-aware dialogues for scalable voice applications.

      • Usage: Automates customer service queries, appointment bookings, and other conversational tasks.

    • Google Contact Center AI (CCAI) Platform

      • Integration: The AI Connect Service connects seamlessly with the CCAI Platform to enhance customer interactions through advanced AI capabilities.

      • Supported Features:

        • Virtual Agents: Utilize Dialogflow CX within the CCAI Platform to build sophisticated virtual agents capable of managing complex dialogues.

        • Agent Assist: Provide human agents with real-time suggestions, content recommendations, and next-best actions during customer interactions to improve efficiency and accuracy.

        • Conversational Insights: Analyze customer interactions for insights such as sentiment trends, call drivers, and agent performance to inform operational improvements.

      • Usage: Automates routine tasks, supports agents in real time, and offers actionable analytics to optimize customer experience and operational efficiency.

  5. Integration with Custom Application Servers

    • REST Callbacks for Metadata: The AI Connect Service integrates with custom applications using REST callbacks to send call metadata such as call IDs, participant information, and events in real time.

    • Real-Time Audio Streaming over WebSocket: AI Connect opens an outgoing connection to a customer-provided WS or WSS endpoint and sends live call audio. Use WSS for TLS-encrypted transport. Audio delivery is separate from the call metadata and event notifications described above. The receiving application performs audio processing or forwards the audio to its chosen AI platform.

    • Channel Selection and Chunk Duration: For SIPRec sessions, streamTarget.channels selects and orders audio channels using channel indices or SIPRec stream labels. streamTarget.chunkDuration configures the audio chunk duration in milliseconds.

    • Authentication: Optional WebSocket handshake headers can be supplied through streamTarget.headers. A Bearer Authorization header matching the /newCall callback configuration is added automatically.

    • Stream Control and Events: Use STARTSTREAM and STOPSTREAM to control audio delivery during an active SIPRec session. STREAMSTARTED, STREAMSTOPPED, and STREAMFAILED report the stream lifecycle. Session and stream identifiers, together with optional streamReferenceId data, let your application correlate these events with the audio. See the SIPRec API for parameters and examples, and Getting Started for the setup flow.

  6. Voice Connectivity

    The AI Connect Service supports integration with telecom systems using industry-standard protocols:

    • SIP Trunking: Facilitates voice stream routing to the analytics engine with minimal latency.

    • SIPRec Protocol: Supplies call audio and metadata for live WebSocket streaming as well as recording workflows used in post-call analysis.

  7. Security Features

    The AI Connect Service incorporates stringent security measures to protect data integrity and confidentiality:

    • Encryption: All communications are secured using TLS for data in transit and AES-256 encryption for data at rest.

    • Access Control: Role-based access is enforced via IAM policies for secure API usage.

    • Data Anonymization: Sensitive information can be anonymized during analysis to ensure privacy compliance.

By leveraging the AI Connect Service's integration capabilities with AWS and Google Cloud services, its robust security features, and the ability to interface with custom application servers, it is a key enabler for businesses wanting to transform voice interactions into actionable insights.